[flang][runtime] Replace recursion with iterative work queue (#137727)

Recursion, both direct and indirect, prevents accurate stack size
calculation at link time for GPU device code. Restructure these
recursive (often mutually so) routines in the Fortran runtime with new
implementations based on an iterative work queue with
suspendable/resumable work tickets: Assign, Initialize, initializeClone,
Finalize, and Destroy.

Default derived type I/O is also recursive, but already disabled. It can
be added to this new framework later if the overall approach succeeds.

Note that derived type FINAL subroutine calls, defined assignments, and
defined I/O procedures all perform callbacks into user code, which may
well reenter the runtime library. This kind of recursion is not handled
by this change, although it may be possible to do so in the future using
thread-local work queues.

The effects of this restructuring on CPU performance are yet to be
measured.

Author: klausler

flang-rt/include/flang-rt/runtime/environment.h

@@ -64,6 +64,9 @@ struct ExecutionEnvironment {
   bool defaultUTF8{false}; // DEFAULT_UTF8
   bool checkPointerDeallocation{true}; // FORT_CHECK_POINTER_DEALLOCATION
 
+  enum InternalDebugging { WorkQueue = 1 };
+  int internalDebugging{0}; // FLANG_RT_DEBUG
+
   // CUDA related variables
   std::size_t cudaStackLimit{0}; // ACC_OFFLOAD_STACK_SIZE
   bool cudaDeviceIsManaged{false}; // NV_CUDAFOR_DEVICE_IS_MANAGED

flang-rt/include/flang-rt/runtime/stat.h

@@ -24,7 +24,7 @@ class Terminator;
 enum Stat {
   StatOk = 0, // required to be zero by Fortran
 
-  // Interoperable STAT= codes
+  // Interoperable STAT= codes (>= 11)
   StatBaseNull = CFI_ERROR_BASE_ADDR_NULL,
   StatBaseNotNull = CFI_ERROR_BASE_ADDR_NOT_NULL,
   StatInvalidElemLen = CFI_INVALID_ELEM_LEN,
@@ -36,7 +36,7 @@ enum Stat {
   StatMemAllocation = CFI_ERROR_MEM_ALLOCATION,
   StatOutOfBounds = CFI_ERROR_OUT_OF_BOUNDS,
 
-  // Standard STAT= values
+  // Standard STAT= values (>= 101)
   StatFailedImage = FORTRAN_RUNTIME_STAT_FAILED_IMAGE,
   StatLocked = FORTRAN_RUNTIME_STAT_LOCKED,
   StatLockedOtherImage = FORTRAN_RUNTIME_STAT_LOCKED_OTHER_IMAGE,
@@ -49,10 +49,14 @@ enum Stat {
   // Additional "processor-defined" STAT= values
   StatInvalidArgumentNumber = FORTRAN_RUNTIME_STAT_INVALID_ARG_NUMBER,
   StatMissingArgument = FORTRAN_RUNTIME_STAT_MISSING_ARG,
-  StatValueTooShort = FORTRAN_RUNTIME_STAT_VALUE_TOO_SHORT,
+  StatValueTooShort = FORTRAN_RUNTIME_STAT_VALUE_TOO_SHORT, // -1
   StatMoveAllocSameAllocatable =
       FORTRAN_RUNTIME_STAT_MOVE_ALLOC_SAME_ALLOCATABLE,
   StatBadPointerDeallocation = FORTRAN_RUNTIME_STAT_BAD_POINTER_DEALLOCATION,
+
+  // Dummy status for work queue continuation, declared here to perhaps
+  // avoid collisions
+  StatContinue = 201
 };
 
 RT_API_ATTRS const char *StatErrorString(int);

flang-rt/include/flang-rt/runtime/type-info.h

@@ -240,6 +240,7 @@ class DerivedType {
   RT_API_ATTRS bool noFinalizationNeeded() const {
     return noFinalizationNeeded_;
   }
+  RT_API_ATTRS bool noDefinedAssignment() const { return noDefinedAssignment_; }
 
   RT_API_ATTRS std::size_t LenParameters() const {
     return lenParameterKind().Elements();
@@ -322,6 +323,7 @@ class DerivedType {
   bool noInitializationNeeded_{false};
   bool noDestructionNeeded_{false};
   bool noFinalizationNeeded_{false};
+  bool noDefinedAssignment_{false};
 };
 
 } // namespace Fortran::runtime::typeInfo